Review of AI in Indian Classical Music

Abstract

The Artificial Intelligence (AI) wave has revolutionised all industries including the music industry. AI is now being leveraged in the music world in ways that could not have been imagined earlier. Music generation, music retrieval, music production are just some of the areas in which AI has started making its mark. There have been significant efforts in literature to review and bring out the impact of AI in Western music. Indian classical music forms are extremely distinct from Western music in terms of ornamentation, emotional depth, and structure. This makes it difficult for the established AI methods in Western music from being leveraged in Indian classical music. Though there have been some efforts made to use AI tools in Indian classical music, there has not been a comprehensive survey of the same. In this paper, we attempt to provide a systematic review of use of AI specifically in the Indian classical music forms of Hindustani and Carnatic music. We cover the three fields of music generation, music retrieval, music production and look into key technologies, datasets and established applications. Additionally, we bring out current challenges, research gaps, and future directions to advance AI-driven innovations in this domain.

Introduction

Introduction

The application of AI in music has grown significantly in recent years, especially in music generation, production, retrieval, and recommendation systems. However, most progress is concentrated in the Western music industry, with limited application to Indian Classical Music (ICM)—which includes Hindustani and Carnatic traditions. ICM is structurally distinct, with intricate elements like raaga, thaala, shruti, and gamakas, making it difficult to directly apply AI models developed for Western music.

This paper provides a systematic review of AI applications in ICM across:

• Music Generation

• Music Retrieval

• Music Production

It surveys 37 papers, highlighting available datasets, APIs, and identifying research gaps and future directions.

2. Indian Classical Music Overview

• Hindustani music (North India) includes forms like Dhrupad, Khayal, Thumri.

• Carnatic music (South India) emphasizes compositions and rhythm.

• Both systems are built on raagas (melodic modes) and thaala (rhythmic cycles).

• Each uses 22 shrutis (microtones) and a 12-note scale.

3. Music Generation

ICM music generation poses challenges due to its emotional depth and improvisational nature. The paper categorizes AI techniques used:

A. Finite State Machines (FSM)

• Used in early approaches for generating simple raagas.

• Limited ability to improvise or scale to complex compositions.

B. Neural Networks

• Generated music based on emotional inputs (e.g., using emotion indices with ChatGPT and neural networks).

C. GANs (Generative Adversarial Networks)

• Used to generate music from MIDI inputs or structured lyrics.

• Struggle with improvisation but generate diverse outputs.

D. Transformers

• Effective for long-term dependencies and polyphonic compositions.

• Require large datasets but well-suited for raaga-based music.

E. Recurrent Neural Networks (RNNs), LSTMs

• Capture long-term patterns.

• Applied for melody generation, emotion-based mapping, and sequence generation.

F. Markov Models

• Used for note prediction with Hidden Markov Models (HMMs).

• Work in two phases: training (learning probabilities) and synthesis (note generation).

Strengths and Weaknesses Summary

4. Music Retrieval

Music Information Retrieval (MIR) in ICM includes raaga, thaala, emotion, and gamaka recognition using AI.

A. Thaala Prediction

• Techniques used: Decision Trees, Random Forests, CNNs, DNNs.

• Achieved accuracies up to 97.5% in Carnatic thaalas using DNNs.

• Some models combine mel-frequency cepstral coefficients (MFCCs) with timbre features.

B. Raaga Identification

• ML and DL methods applied: KNN, SVM, Random Forest, CNNs, DNNs, LSTMs.

• High accuracies (e.g., 99.72% for Melakarta raagas using Random Forest).

• Novel work included explainable AI models (SoundLIME, GradCAM++) to match human understanding of raagas.

C. Emotion, Gamaka, and Genre Detection

• Emotions (rasa) mapped to music using ML/DL.

• Gamaka identification (ornamentations) done with spectral and temporal analysis.

• Genre classification explored with CNNs and audio descriptors.

5. Key Contributions of the Paper

1. Comprehensive Framework – Categorizes AI technologies for ICM in generation, production, and retrieval.

2. Literature Survey – Covers 37 papers, tools, datasets, and APIs relevant to ICM.

3. Impact Analysis – Evaluates AI's practical contributions to preserving and enhancing Indian music.

Conclusion

In this review paper, we have attempted to survey important research in the domain of Indian classical music. We have brought out the advances made in literature in the fields of music generation, music retrieval and music production. We have also brought out the tools and open-source dataset for Carnatic and Hindustani music families. Additionally, we have also highlighted the challenges faced by AI in ICM. We have endeavoured to make it a comprehensive survey which provides any beginner in the field, an overall knowledge of the developments in the field. There is a lot of work which needs to be done in terms of music production for Indian classical music. Through this review, we have also been able to brought out that there are rich resources in terms of datasets and APIs however, there is a pressing need for larger volume of data with richer characteristics covering further gharanas (sub-genres) in ICM. Further, most of the work in terms of instrumentation are captured in terms of Tabla, Sarod etc. ICM boasts of a plethora of rich instruments of flutes, veenas, sitar, shenai etc. Digitalisation of historically recorded audio tracks and making them available as publicly annotated datasets will strengthen AI research in the domain. Lastly, most of the work in ICM was found in literature alone. Very few supporting codes could be found on publicly hosted sites such as Github. Public sharing of codes and generated models needs to be encouraged to further advance research in this domain. We anticipate that future researches would be directed towards evolving AI from standalone/piecemeal system to a comprehensive end-to-end music aide for an immersive music experience.

References

[1] Adhikary, S and M, Manasa S. and K, Shreesha S L, Kamatchi Priya,. \"Automatic Music Generation of Indian Classical music based on raga.\" IEEE 8th International Conference for Convergence in Technology. 2023. 1-7. [2] Agwan, M. and M. N. \"The Fusion of AI and Music Generation: A Comprehensive Review.\" 6th International Conference on Advances in Science and Technology (ICAST). 2023. 90-94. [3] Anand, A. \" \"Raga Identification Using Convolutional Neural Network.\" Second International Conference on Advanced Computational and Communication Paradigms (ICACCP. 2019. [4] Aswale, S. and Shrivastava, P.C. and Ranjan, R and Shende, S. \"Indian Classical music Recognition using Deep Convolution Neural Network.\" International Journal of Electrical and Electronics Research, 2024: 73-82. [5] Basu, M. and Sanyal,S. and Banerjee,A. and Nag,S and Banerjee,K. and Ghosh,D. \"Neural quantification of timbre and emotions from Indian Classical music: A multifractal exploration,.\" Physica A: Statistical Mechanics and its Applications,, 2023. [6] Bellur, S. and Hong,A. and Ma, J.S. \"An Ensemble Approach to Raga Classification using Deep and Traditional Machine Learning.\" IEEE 16th International Conference on Computational Intelligence and Communication Network. 2024. 195-199. [7] Bhagvathi, A. a. (2019). ragamAI: A Network Based Recommender System to arrange an Indian Classical music Concert. 18th IEEE International Conference on Machine Learning and Applications. [8] Br, Atreya and Haldar, A. and Biswas, A. and Ganguly, S.N. \"From Raaha to Notation: Bridging emotions and music through Neural Networks.\" International Journal of Science and Research, 2022. [9] Carnatic Music Junior Grade Text Book. Bangalore: Managing Director, Karnataka Text Book Society (R), Govt of Karnataka, 100 Feet Ring Road, BSK III Stage, Bangalore – 560085. [10] Chakraborty, S., Tewari, S., Rahman, A., Jamal, M., Lipi, A., Chakraborty, A., et al. (2021). Hindustani Classical music: A Historical and Computational Study. Sanctum Books. [11] Churi, P and Patil, Y. and Patil, S. and Patil, S. and Magar, A. \"Emotion based Music Generation based on Nava Rasas in Indian Classical music.\" Ijraset Journal For Research in Applied Science and Engineering Technology, 2023. [12] CompMusic. (2023). Indian art music tonic datasets. [13] Deb, A and Mahapatra, A.S. and Sanyal, S. and Banerjee,A. and Nag, S. and Basu, M. and Ghosh, D. \"Emotion and Instrument Recognition from Indian Classical music Recordings Using Transformers.\" Proceedings of 27th International Symposium on Frontiers of Research in Speech and Music, 2024. [14] Deolekar, S and Abraham, S. \"Tree based classification of Tabla Strokes.\" Current Science, 2018: 1724-1731. [15] Desai, B U and Tandel, K.I, and Patel, R.M. \"Finite State Machine for Artificially Generating Indian Classical music.\" International Research Journal of Engineering and Technology, 2020: 7658-7661. [16] Deshmukh, S and Bhirud, S. \" North Indian classical music’s singer identification by timbre recognition using MIR toolbox.\" International Journal of Computer Applications , 2014. [17] Gopi, S. and William, F. \"Introductory Studies on Raga Multi-track Music Generation of Indian Classical music using AI.\" AIMC. UK, 2023. [18] Gupta, K. \"Improvised Sequence Generation in North Indian Classical music.\" 2022. [19] Hettiarachchi, H.B.N and Lekamge, L.S and Charles, J and Lekamge L.S. \"Taal Recognition of North Indian Classical music: A Data Mining Approach.\" International Conference on Advances in Computing and Technology. 2020. [20] \"hindustaniclassical.com.\" History of Indian Classical music. [21] Humse, K. ., H S, R. K. ., V, V., K, R., Ramachandra, L. S. ., G R, Y., Puttegowda, K. ., & R, B. . \" Automated raga recognition in Indian classical music using machine learning techniques.\" Journal of Integrated Science and Technology, 13(1),, 2024. [22] Jayakumar, A., Pai, M.L. \" Taala Classification in Carnatic Music Using Machine Learning Algorithms and Deep Neural Networks.\" Advanced Computing and Intelligent Technologies. Lecture Notes in Networks and Systems, 2022. [23] Jayatharan, V and Alwis, D. \"Alapana Generation using Finite Sttate Machine and Generative Adversarial Network.\" International Confeence on Soft Computing and Software engineering. 2023. [24] Joshi, D and Pareek, J and Ambatkar, P. \"Indian Classical Raga Identification using Machine Learning.\" International Semantic Intelligence Conference. 2021. 259-263. [25] Kharvatkar, S.R. and Sharma,M. and Khairnar, D.G. and Naik, I.C. \"Detection of Pitch Frequency of Indian Classical music Based on Hilbert-Huang Transform for Automatic Note Transcription.\" Fourth International Conference on Computing Communication Control and Automation. 2018. [26] Konduri, S. a. (2024). KritiSamhita: South Indian Music Tonic Recognition Dataset. [27] Krishnan, V. a. (2025). Sanidha:A studio quality multi-modal dataset for carnatic music. arxiv. [28] Lele, J. and Abhyankar, A. \"Indian Music Generation and Analysis using LSTM.\" Jounral of Propulsion Technology, 2024. [29] M. Vyas, Suma S. M., S. G. Koolagudi and Guruprasad K. R. \"Identifying gamakas in Carnatic music.\" Eighth International Conference on Contemporary Computing. 2015. [30] Majumdar, S. a. (2024). Thaat and Raga Forest(TRF) Dataset. [31] Maximos Kaliakatsos-Papakostas,, A. (2020). Artificial intelligence methods for music generation: a review and future perspectives. Nature-Inspired Computation and Swarm Intelligence , 217-245. [32] Mehta, A and Chauhan S, and Choudhury, M. \"Missing Melodies: AI Music Generation and its \"Nearly\" Complete Omission of the Global South.\" arXiv. 2024. [33] Miguel Civit, J. C.-M. \"A systematic review of artificial intelligence-based music generation: Scope, applications, and future trends.\" Expert Systems with Applications , 2022. [34] Mishra, P. and Kalra, M. \"Time-Specific Classification in Hindustani Classical music,.\" 5th International Conference for Emerging Technologies. 2024. [35] Misra, C. a. (2023). Sangeet: A XML based Open Dataset for research in Hindustani Sangeet. arXiv . [36] Mycka. \"Artificial intelligence in music: recent trends and challenges.\" Expert Systems and Int, 2024. [37] Mycka, J. M. (2024). Artificial intelligence in music: recent trends and challenges. Neural Computing and Applications . [38] Nag, S. and Basu,M. and Sanyal,S. and Banerjee,A. and Ghosh,D. \"On the application of deep learning and multifractal techniques to classify emotions and instruments using Indian Classical music.\" Physica A: Statistical Mechanics and its Applications,, 2022. [39] Nayak, S. D S and J. S. \"Machine Learning based Classification and Recognition of Melakarta Ragas using Melakarta Chakra Analysis.\" 8th International Conference on Computational System and Information Technology for Sustainable Solutions. 2024. [40] Nicholls, S and Cunningham S, and Picking,R. \"Collaborative Artificial Intelligence in Music Production.\" Proceedings of Audio Mostly 2018: Sound in Immersion and Emotion. 2018. [41] Patil, S. T. Patil and Abhay,P.P, and Prakash,L.S. and Arun, K.M and Arun, P.A. \"Classification of Classical Indian Music Tabla Taals using Deep Learning.\" International Journal on Recent and Innovation Trends in Computing and Communication, 2024: 3227–3232. [42] Pawar, R. and Gujar, S. and Bidkar, A. and Dandawate, Y. \"Classification of Hindustani Classical Ragas unsing one-dimensional convolution neural networks.\" Big data, machine leanring and applications , 2024. [43] Porter, A. a. (2013). Dunya: A System for Browsing Audio Music Collections Exploiting Cultural Context . [44] Ragesh, R.M. and Vijayasenan, D and Vijaykumar, A. \"Predicting Gamakas-the essential embellishments in Karnatic music.\" IEEE Access, 2019. [45] Rao, B. C. and Arkachaari, V. and Harsha, M N and Sushmitha, M N* and Gayathri, Ramesh K K* and Ullas, M S*. \"Bi-Sampling Approach to Classify Music Mood leveraging Raga-Rasa Association in Indian Classical music.\" arXiv, 2022. [46] Rickard, Ebba. \"Generating Music using AI.\" 2022. [47] Roberts, A. and Engel,J. and Raffel, C. and Hawthorne, C and Eck,D. \"A Hierarchical Latent Vector Model for learning long term structure in Music.\" International Conference on Machine Learning. 2018. 4364-4373. [48] Saxena, D., Saxena, D., Saraf, L., Saxena, L., Banerjee, D., Chaudhary, D. R., et al. (2019). Hindustani Music Theory. Secretary, National Institute of Open Schooling. [49] Schedl, Markus and Gómez, Emilia and Urbano, Julián. \"Music Information Retrieval: Recent Developments and Applications.\" 2014. [50] Sharma, A.K. and Aggarwal, G. and Bhardwaj, S. and Chakrabarti, P. \"Classification of Indian Classical music with Time-Series Matching using Deep Learning.\" IEEE Access, 2021. [51] Shete, S. and Deshmukh, S. (2021). \"North Indian Classical music Tabla Tala (Rhythm) Prediction System Using Machine Learning.\" Advances in Speech and Music Technology. Advances in Intelligent Systems and Computing, 2021. [52] Simonetta, F and Carnovalini, F. and Orio N. and Roda, A. \"Symbolic music similarity thorugh a graph based representation.\" proceedings of the Audio Mostly 2018 on Sound in Immersion and Emotion. 2018. 1-7. [53] Singh, P and Arora, V. \"Explainable Deep Learning Analysis for Raga Identification in Indian Classical music.\" ArXiv, 2024. [54] Singh, Y. and Waikhom, L. and Biswas, A. \"Swaragram: A software toolbox for musical feature of indian music.\" Software Impacts, 2023. [55] Singha, A. and Rajalakshmi, N.R. and Pandian, J. A. and Swaminathan, S. \"Deep LEarning Based Classification of Indian Classical music based on Ragas.\" Sizth International Conference on Information systems and Computer Networks. 2023. 1-7. [56] Streich, G and Lanzendorfer, L.A. and Grotschla, F. and Wattenhofer, R. \"Generating vocals from Lyrics and Musical Accompaniment.\" Audio Imagination:NeurIPS 2024 Workshop AI-Driven Speech, Music and Sound Generation. 2024. [57] Srinivasamurthy, A. a. (2021). Saraga: Open datasets for research on indian art music. Empirical Musicology review, , 85-98. [58] Surana, R. a. (2022). Deep Learning for Conversions Between Melodic Frameworks of Indian Classical music. Proceedings of Second International Conference on Advances in Computer Engineering and Communication Systems. Algorithms for Intelligent Systems. [59] Sutar, P and Saitwal, M. and Wakchaure, S and Shinde, R and Sawdekar, P. \"A Survey on Raga Identification using Machine Learning Techniques.\" INTERNATIONAL JOURNAL OF INNOVATIVE RESEARCH IN TECHNOLOGY , 2020. [60] Tiple, B. and Patwardhan, M. \"Multi-label emotion recognition from Indian classical music using gradient descent SNN model.\" Multimed Tools Appl 81, 2022: 8853–8870. [61] Varadharajan, J. S. (2014). Automatic Synthesis of Notes Based on Carnatic Music Raga Characteristics. Advanced Computing, Networking and Informatics- Volume 1. Smart Innovation, Systems and Technologies, vol 27. [62] Vidwans, A. and Rao, P. \"Identifying Indian Classical music Styles using Melodic Contours.\" FRSM. 2012. [63] Viramgami, G. and Gandhi, H and Naik, H and Mahajan, N. \"Indian Classical music Synthesis.\" Proceedings of the 5th joint international conference on data science and management of data. 2022. 322-323. [64] Muñoz Lago, Paula –“A Deep Learning Approach to generate Beethoven\'s 10th Symphony”, 2019. [65] A. Das, M. Choudhury, and S. Sarkar, \"Finite state models for generation of Hindustani classical music,\" Proceedings of the Workshop on Constraints and Language Processing (CSLP-05), 2005 [66] www.wikipedia/Carnatic music.

Copyright

Copyright © 2025 Rachel Vinod, Robin Colaco. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.